Agricultural activities are the main livelihood for about 70% of Tanzania's population, with women being the main players. Crops need water (crop water requirements, CWRs) for their growth and production, which can either be rain-fed or irrigation sourced. However, climate change has affected the hydrological cycle, particularly water available for agricultural crops. Since impacts and consequently adaptation are site-specific, an assessment of the effects of climate change on maize water requirements in Kikafu sub-catchment was conducted using a crop simulation model, CROPWAT. Accordingly, climate scenarios were obtained from A2 emission scenario using three general circulation models (GCMs). These scenarios were downscaled at two site locations using the Long Ashton Research Station Weather Generator (LARS-WG) model. The baseline period for the change analysis was 1971–2000. The CWRs are projected to increase by 3.8% in the 2020s and 7.1% in the 2050s at the Moshi Airport and 19.9 and 22.4% at Lyamungu station, respectively. More impact is projected to be during 70–80 days of the development stage and the entire mid-season (81–140 days) whereby the temperature will be high but with low precipitation. With the increasing CWRs, better adaptation measures are increase crop diversification, restore soil organic matter and change cropping systems as established through the multi-criteria analysis.

INTRODUCTION

The ever changing climate constitutes a major challenge to mankind today. Various sectors of the economy, such as agriculture, energy, tourism, are increasingly being affected by climate change. Water and food security are under severe threat escalated by increased world population and food demand. Consequently, most of the water resources are being depleted and the agricultural production decreasing with the reduction in arable lands. On the whole, climate change is a global problem with its effects being severe in developing countries where the majority of the people are poor and depend primarily on rain-fed agriculture (Morton 2007). As the bulk of the populations in Africa depend on rain-fed agriculture for food and their livelihoods, they are vulnerable to the climate effects. Climatic data indicate that the continent experiences decreasing and increasing trends of rainfall and temperature, respectively (IPCC 2007). Studies have predicted that the average global temperature may increase by 1.4–5.8°C and there would be a substantial reduction in freshwater resources and agricultural yield by the end of the twenty-first century (Tadross & Wolski 2010). Moreover, climate change has resulted in an increase in globally averaged mean annual air temperature and variations in regional precipitation, with these changes projected to continue intensifying in future (Solomon et al. 2007). Increased temperatures lead to increased evapotranspiration, which affects water availability for crops (Holmén 2003). In some African regions, the projected climate change is likely to increase water stress and demand for increased irrigation, hence leading to a drop in crop yields (Ravindranath et al. 2010). Projections, which show a general increase in average temperatures for Africa, are predicted to result in increased rainfall variability and incidences of extreme weather conditions (AUC-AMCOW 2016) such as floods and droughts. Changing rainfall patterns will in turn negatively affect cropping systems (AUC-AMCOW 2016). Droughts and floods cause a great deal of failures and damage to the crops, and thus result in food shortages (Liwenga et al. 2007). Also, changes in rainfall patterns and amounts have led to a loss of crops (Rosenzweig et al. 2002). However, developing countries, most of them in Africa, lag behind, particularly in climate downscaling studies (Joubert & Hewitson 1997; Souvignet et al. 2010) whereas climate risks are high due to low resilience and adaptation capacity for these countries. Therefore, there is a need for adaptation measures to be in place to contain the worrisome situation. In simple terms, adaptation is about changing behaviour and practices in response to a certain risk or impact.

In Tanzania, the economic base is largely dependent on natural resources, rain-fed agriculture, and biomass for household energy which are highly vulnerable to adverse climate change impacts (NAPA 2006). The country has been experiencing increases in temperature which have led to devastating droughts and which contribute to power crises, decreases in lake levels, lake recessions and melting of glaciers atop Mount Kilimanjaro. NAPA (2006) also reported that climate change is also impacting on the country's economic development with the GDP growth rate falling by 0.1% in 2005. This reduction was attributed to the severe drought in most parts of the country that triggered food shortages and power crises. As agriculture in Tanzania is mostly rain-fed, the success of activities in the sector remains highly sensitive to weather, especially rainfall. Agriculture is the largest sector in Tanzania's economy, which contributes 50% of the GDP and exports. About 80% of the population live in rural areas and rely primarily on agriculture with the sale of the agricultural produce contributing 70% of rural household incomes (URT 2006). The agricultural production in Kikafu River sub-catchment in the Pangani River basin is mainly rain-fed. As a result, there was a need to assess the impacts of climate change on crop water requirements (CWRs) and work on adaptation measures to alleviate the impact. Also, the sustainable development goal 13: Climate Action, emphasises taking urgent action at all levels to combat climate change and its impact. To date, little is known about climate downscaling and impact studies in Tanzania (Gulacha & Mulungu 2016), and for the study area in particular. The downscaled data is important for studies on impacts of climate change that are spatially distributed. In this case, Kikafu sub-catchment was considered as a case study with the Long Ashton Research Station Weather Generator (LARS-WG) model applied for downscaling of the climatic data. The assessment of the effects of climate changes on agriculture may help to anticipate and appropriately adapt the effects of farming in order to sustain agricultural production. Farming in Kikafu sub-basin in the Kilimanjaro area involves mainly rain-fed agriculture and is, therefore, vulnerable to climate perturbation. To support farming under anticipated climate change, this study worked on future climate change scenarios, determined impacts of climate change on CWRs and proposed adaptation strategies for Kikafu sub-basin. The study focuses on two major variables of the climate: precipitation as water supply (crop water supply) and temperature, which influences evapotranspiration (crop water demand). Surface (screen height) observations of air temperature and precipitation are the most common variables applied in impact studies (IPCC-TGICA 2007).

From an agroclimatological point of view, simple changes in temperature, precipitation and moisture can restrict agricultural output potential and affect crop growth in the process (Noufea et al. 2015). In addition to the above climatic variables, solar radiation, humidity and wind speed were used in this study for CROPWAT modelling to analyse the changes in CWRs for a maize crop. The impact of climate change on CWRs can be an important indicator of changes in the capacity for crop production in the sub-catchment. Noufea et al. (2015)’s study found a generally high correlation coefficients between crop water requirements index (CWRI) and crop (maize) yields for the dry and wet conditions among agroclimatic indicators (satisfaction of the CWRI, duration of cropping season, onset of the cropping season, end of the cropping season). Also, to some extent, there was a high correlation between duration and maize yields. However, that study cautioned on the quality of the data used, particularly the yields data. Accordingly, Noufea et al.’s (2015) study highlighted the value of analysing impacts of climate change on CWRs. Rehana & Mujumdar (2013) dealt with impacts of climate change on CWRs and subsequent irrigation water demand using the crop coefficient approach for A1B scenario in India for paddy, sugarcane, permanent garden and semi-dry crops. The work used only one climate scenario; in the meantime, little is known about the impacts and adaptation on other scenarios and other crops such as maize. Maize as a rain-fed crop is likely to be impacted by warming, reduction in rainfall and water stress (Ravindranath et al. 2010). Therefore, the objective of this study was to assess the impacts of climate change on maize CWRs and analyse adaptation strategies that can be used to overcome or minimise the adverse impacts on maize crop in Kikafu sub-catchment.

METHODS

Study area description

The Kikafu River sub-catchment is based in the north east of Tanzania. It is located in the Pangani River basin, which covers an area of 43,650 km2 with 95% of the area in Tanzania and 5% in Kenya. The sub-catchment is also located in the Kilimanjaro region between latitude 3°02′40″–3°20′22″ east and longitude 37°09′56″–37°19′49″ south and has a catchment area of about 208 km2 (Figure 1). Kikafu River is one of the rivers that drain into Kikuletwa River. The Kikuletwa River drains the slope of Mount Meru and the southern slope of Mt. Kilimanjaro.
Figure 1

Kikafu River sub catchment in Pangani River basin.

Figure 1

Kikafu River sub catchment in Pangani River basin.

The elevation in the sub-catchment lies between 977 and 5,267 metres above sea level (masl), whereas the Uhuru peak of Mt. Kilimanjaro is at 5,892 masl and the Moshi Airport and Lyamungu weather stations are, respectively, at 813 and 1,250 masl. Rainfall patterns are largely related to altitude, with the highlands receiving about 1,000–2,000 mm annually, and the lowlands receiving 500–600 mm. Rainfall is bimodal, with the rainy season occurring mainly in March to June, with the short rains from November to December. Human settlements and agricultural activities are located in the plains and the lower slopes of Mount Kilimanjaro. Kikafu sub-catchment has about 43% inhabited area whereas 57% of the area is still virgin land (Rohr 2003). It is characterised by small-scale irrigated agriculture. The main crops grown are maize, coffee, beans and bananas (IUCN 2010).

Datasets

Agroclimatic data

The SSI-2 project (Mulungu 2014) provided daily climatic data for the study (Table 1), which were checked, corrected and processed in time series format, particularly climatic data up to 2011 for Moshi Airport and Lyamungu stations. The more recent data for these stations could not be obtained.

Table 1

Available climatic data at Moshi Airport and Lyamungu stations

Station code Station name Data type Records Missing data (%) 
09337004 Moshi Airport Rainfall 1929–2011 2.30 
09337021 Lyamungu Rainfall 1935–2011 2.68 
09337004 Moshi Airport Minimum temperature 1958–2011 0.03 
09337004 Moshi Airport Maximum temperature 1958–2011 0.07 
09337021 Lyamungu Minimum temperature 1962–2011 0.00 
09337021 Lyamungu Maximum temperature 1962–2011 0.00 
09337004 Moshi Airport Relative humidity 1972–2004 21.91 
09337021 Lyamungu Relative humidity – – 
09337004 Moshi Airport Sunshine hours 1958–2011 0.00 
09337021 Lyamungu Sunshine hours 2001–2011 0.00 
09337004 Moshi Airport Solar radiation Estimated 0.00 
09337021 Lyamungu Solar radiation Estimated 0.00 
09337004 Moshi Airport Wind speed 1980–2007 2.76 
09337021 Lyamungu Wind speed – – 
Station code Station name Data type Records Missing data (%) 
09337004 Moshi Airport Rainfall 1929–2011 2.30 
09337021 Lyamungu Rainfall 1935–2011 2.68 
09337004 Moshi Airport Minimum temperature 1958–2011 0.03 
09337004 Moshi Airport Maximum temperature 1958–2011 0.07 
09337021 Lyamungu Minimum temperature 1962–2011 0.00 
09337021 Lyamungu Maximum temperature 1962–2011 0.00 
09337004 Moshi Airport Relative humidity 1972–2004 21.91 
09337021 Lyamungu Relative humidity – – 
09337004 Moshi Airport Sunshine hours 1958–2011 0.00 
09337021 Lyamungu Sunshine hours 2001–2011 0.00 
09337004 Moshi Airport Solar radiation Estimated 0.00 
09337021 Lyamungu Solar radiation Estimated 0.00 
09337004 Moshi Airport Wind speed 1980–2007 2.76 
09337021 Lyamungu Wind speed – – 

In Kikafu sub-catchment, agricultural production is mainly carried out by smallholder farmers belonging to self-sustaining households growing a variety of crops and in mixed cropping, with maize and beans being the most common crops. The assumption was that there are no changes in the cropping season, soil conditions and crop varieties. Crop data were required and used in the crop simulation model, CROPWAT. The CROPWAT model requires climatic parameters (e.g. maximum and minimum temperatures wind speed, sunshine, humidity and rainfall), planting and harvesting dates of crops, soil type and cultivation area of crops for estimating CWR. The missing climatic data were provided in the model (Allen et al. 1998; Savva & Frenken 2002). The crop coefficient values for the maize crop considered (see Table 2) were obtained from FAO (1998, 2015). The planting and harvest dates for maize were 3rd March and 29th August, respectively. This maize variety has a growing season of 180 days (Table 3).

Table 2

Crop coefficients, Kc values for maize (FAO 1998, 2015)

Initial season Development and mid season Late season 
0.70 1.20 0.41 
Initial season Development and mid season Late season 
0.70 1.20 0.41 
Table 3

Planting date, length of growing stages (days) and harvest date for maize

  Maize Crop growing stages (days)
 
  
Planting date Initial Development Mid-season Late-season Harvesting 
3 March 30 50 60 40 29 August 
  Maize Crop growing stages (days)
 
  
Planting date Initial Development Mid-season Late-season Harvesting 
3 March 30 50 60 40 29 August 

Climatic data from general circulation models

The general circulation models (GCMs) were used to obtain future climate and climate change scenarios. The GCMs are computer-based models designed to simulate historical and future climate based on greenhouse gases emission scenario inputs to project potential climate scenarios. They provide geographically and physically consistent estimates of regional climate change, which are required in impact analyses. They are fully coupled with mathematical representations of the complex physical laws and interactions between ocean/atmosphere/sea-ice/land-surface (Smith & Hulme 1998). As the different GCMs are run by different centres, differences in prediction are bound to happen. Therefore, selection of the appropriate GCM for use can be difficult, especially when more than one is available. The criteria for the GCM selection involve considering the resolution, vintage and validity (Smith & Hulme 1998). Also, Tubiello et al. (2000) recommend using different GCMs in impact studies aimed to assess climate change on crop yields. To achieve representativeness of the results, the results should be drawn from more than one GCM in an impact assessment (IPCC-TGICA 2007). For example, a study in southern Africa adopted three GCMs (Hulme 1996). Accordingly, three GCMs (Geophysical Fluid Dynamics Laboratory (GFDL) in the USA, Centre National de Recherches Meteorologigues (CMCM3) in France and Hadley Centre for Climate Prediction and Research (HADCM3)) were selected and used in the study. These GCMs were used in international model intercomparison projects and are relevant to impact assessment (IPCC-TGICA 2007). In this case, the spatial variability in climatic variables between the Moshi Airport and Lyamungu stations was captured using data observed at the stations to derive the site's parameters, and by developing relationship between the large-scale GCM (predictor) and local surface variables (predictand variables of the rainfall and temperature observed) through the downscaling model, LARS-WG. The site's parameters include semi-empirical distributions for length of dry and wet series, precipitation, minimum and maximum temperature and radiation calculated separately for dry and wet days and correlation and auto-correlation coefficients for each month.

This study used the IPCC AR4 (Fourth Assessment Report) scenario data as during the time of research, 2013–2014, the IPCC AR5 (Fifth Assessment Report) data could not be accessed. As a result, the A2 emission scenario was selected for the study. This A2 scenario was deemed relevant for application in developing countries. To begin with, the A2 scenario describes a very heterogeneous world, whereby the underlying theme is self-reliance and preservation of local identities, with higher population, slow and fragmented economic growth and technological changes. More significantly, the scenario emphasises on regional, local, social, economic development aspects.

Downscaling of climatic data

LARS-WG model

The GCM output is rather coarse, typically in the order of 2–3 degrees. The coarse spatial resolution is a problem in direct application of projections on local assessment of climate change impacts. This coarse resolution results in significant errors, biases and large uncertainty in their output at a local scale precipitation (Iizumi et al. 2010; Eden et al. 2012). Due to the coarse spatial resolution, downscaling helps to obtain finer local scale spatial resolution necessary in impact analysis. The downscaling techniques include dynamic downscaling and statistical downscaling and weather generators.

The statistical downscaling was used as it requires less computational effort which offers an opportunity for testing scenarios for many decades. It requires the existence of a relationship between the predictors and the predictands (Wilby et al. 2002). It applies the information from GCMs to the region by using a series of equations to relate variations in global climate to variations in local climate (Wilby et al. 2002). Accordingly, the LARS-WG was used in this study as it is a computationally inexpensive tool in producing site-specific climate scenarios for climate change impacts (Semenov 2009). The LARS-WG is a stochastic weather generator used for simulating weather data at a single site under both current and future conditions (Semenov et al. 1998). It uses weather data observed on a daily basis for a given site to compute a set of parameters for probability distributions of weather variables as well as correlation between them, which are used to generate synthetic weather time series of arbitrary length by randomly selecting values from the appropriate distributions. Also, LARS-WG generates future climate scenarios by altering the baseline site parameters using change factors derived from climate projections (Semenov & Stratonovitch 2010; Iizumi et al. 2012). By using change factors derived from climate projections to perturb parameters of site distributions of climatic variables, LARS-WG can generate plausible future climate scenarios at a site with weather statistics similar to those predicted by climate models (Semenov & Stratonovitch 2010). LARS-WG has been intensively tested over different climates and its performance in representing climate statistics observed was generally good and the overall performance of LARS-WG in representing the statistical characteristics of climatic variables observed, including extreme events, was generally good (Semenov et al. 1998; Iizumi et al. 2012). In this study, LARS-WG 5.5 was used. This version has been incorporated with predictions from IPCC AR4 (Fourth Assessment Report) multi-model ensemble.

The conceptual framework for LARS-WG application including calibration is as presented in Figure 2, which was used to generate climate change scenario data.
Figure 2

The LARS-WG framework of generating local-scale daily climate change scenario data.

Figure 2

The LARS-WG framework of generating local-scale daily climate change scenario data.

From the weather data for a site observed daily, site analysis is undertaken to compute site parameters. This is the same as model calibration. The site parameters may be refined by changing the random seed number. The site parameters obtained are used to generate synthetic daily weather data for a site, which statistically resembles observed weather. The performance of calibrated LARS-WG model is tested by three statistical tests along with corresponding p-values, which assess its ability to reproduce variety of weather statistics accurately: the Kolmogorov-Smirnov (K–S) test, which compares the probability distributions; the t-test, which compares means and the F-test, which compares standard deviations derived from the generated and weather observed. The p-value indicates the likelihood that generated and data observed originate from the same distribution. If the p-value is very low and below the significance level (0.05), then the generated simulated climate is unlikely to be the same as the ‘true’ climate. By applying changes to the climate derived from a global or regional climate model, LARS-WG can generate site-specific daily weather.

Generation of climate change scenario at the local level

To generate future climate scenarios, the baseline site parameters are adjusted by the changes from the future time period obtained by using the A2 scenario as predicted by the GCMs and downscaled by LARS-WG. Changes in the mean temperature and solar radiation are additive values whereas changes in monthly precipitation, length of wet and dry spells and temperature standard deviation are multiplicative. The changes are calculated as relative changes for precipitation and radiation as they are dependent on the previous climatic condition whereas for minimum and maximum temperature they are calculated as absolute changes as they are not dependent on the previous climatic condition.

LARS-WG uses sets of ratios and differences to perturb the site parameters to obtain the daily climate scenario for a changed climate (Semenov & Barrow 2002). The multiplicative values are obtained from the ratio of the future climate data to the baseline climate data whereas additive values are obtained from the subtraction of baseline data from the future climate data. By perturbing parameters of distributions for a site with the predicted changes of climate derived from global or regional climate models, a daily climate scenario for the site could be generated and used in conjunction with CROPWAT, which is a crop simulation model for the assessment of impacts (Semenov & Stratonovitch 2010). In this study, the local-scale climate scenarios based on the A2 scenario simulated by the selected three GCMs were generated using LARS-WG 5.5 for the time periods of 2011–2030 and 2046–2065 to predict the future change in climate and impacts on monthly and decadal CWRs. The above time periods were named as 2020s or tier one (T1) and 2050s or tier two (T2) respectively.

Climate change impacts on CWRs

Trend analysis of rainfall and temperature

Local or site air temperature data can provide a good indicator for the evapotranspiration (ET) or evaporation (E) that is atmospheric driven due to the level of warmth on the land that can influence the rates of ET or E locally. Also, for rain-fed agriculture, rainfall is the only water supply for crops. These two climatic variables were, therefore, used for the trend analysis to indicate the historical variations of the climate, thus confirming the changing climate. In this case, the baseline data for the period 1971–2000 was analysed to determine the existence of linear trend. The trend analysis was carried out using non-parametric Mann–Kendall test (WMO 1988; Kundzewicz & Robson 2000; Adnan & Atkinson 2011) to detect the pattern change in the precipitation and temperature data and identified whether those changes were significant or not. Trend magnitude was evaluated using the non-parametric Sen's slope approach (Hirsch et al. 1982; Tan et al. 2015).

Analysis of climate change impacts

A comparison of CWR patterns during baseline and future scenario periods was made to analyse the climate change impacts on CWRs. The analysis looked at satisfaction or meeting of CWRs for the growth stages and for the whole growing period. The satisfaction is considered when the crop evapotranspiration (ETc or AET) is less or equal to the effective rainfall. The analysis of impacts considered a fixed cropping season with the assumption of no changes in soil conditions and maize crop variety. In this case, a crop simulation model, CROPWAT was used. CROPWAT is a collection of modules following the Penman–Monteith method that integrates several models necessary to predict CWR, irrigation water management and crop scheduling (Smith 1991). The Food and Agriculture Organization (FAO) approved the Penman–Monteith method to predict reference evapotranspiration (ETo), CWRs and irrigation water management (Smith 1991; FAO 1998). The Penman–Monteith method has been recommended by the FAO for its appropriate combinations of relevant climatic parameters for predicting ETo (FAO 1998; Smith & Kivumbi 2006). The FAO Penman–Monteith method is the sole recommended method for determining reference crop evapotranspiration (ETo) (Equation (1)) and it provides ETo values that are more consistent with actual crop water use data in all regions and climates (Allen et al. 1998; Savva & Frenken 2002). The method has been proved to have global validity as a standardised reference for grass evapotranspiration and has found recognition by the International Commission for Irrigation and Drainage, by the World Meteorological Organization as well as by a large number of scientific studies (Allen et al. 1998; FAO 2016). The ETo, which is the amount of water evaporated from both the plant and soil surface was calculated using Equation (1) in the CROPWAT model. Due to the limited availability of all the climatic data, this method uses measured minimum and maximum temperature data as well as general levels of humidity, sunshine and wind, to calculate ETo: 
formula
1
where ETo is reference evapotranspiration (mmday–1); Rn is net radiation at the crop surface (MJ m–2 day–1); G is soil heat flux density (MJ m–2 day–1); T is mean daily air temperature at 2 m height (°C); u2 is wind speed at 2 m height (m s–1); es is saturation vapour pressure (kPa); ea is actual vapour pressure (kPa); esea is saturation vapour pressure deficit (kPa); Δ is slope vapour pressure curve (kPa °C–1); γ is psychometric constant (kPa °C–1).
The CROPWAT calculates the CWR or ETc (mm day–1) using the equation: 
formula
2
where ETo = reference evapotranspiration (mmday–1); Kc = crop coefficient at a specific growth stage.

The Kc depends on the type of crop (e.g. height of crop, resistance of canopy, albedo), soil and climatic parameters, such as soil surface, evaporation and wind speed and direction (FAO 1998). Albedo is the fraction of solar radiation reflected by the surface of crop and soil whereas the canopy refers to the leaves and branches of crops that make a kind of roof. Resistance of canopy is the resistance of the crop against vapour transfer (FAO 1998). Kc varies depending on the type of the crop and the growing stage of a crop (e.g. initial stage, crop development, mid-season and late season) as indicated in Table 2. The Kc approach enables the obtaining of the actual evapotranspiration, which is the CWR. The Kc varies predominately with the specific crop characteristics and – only to a limited extent – with climate. This enables the transfer of standard values for Kc between locations and between climates. This has been the main reason for the worldwide (global) acceptance and usefulness of the crop coefficient approach and the use of Kc factors developed in previous studies (Allen et al. 1998; Savva & Frenken 2002).

Multi-criteria analysis of adaptation strategies

The analysis of climate change using precipitation and temperature variables showed that the Kikafu area is vulnerable to climate change impacts on CWRs. The impacts were then addressed in the form of adaptation strategies as a long-term measure. The climate change impact on maize crop may be increased or decreased CWRs. Accordingly, the adaptation measures were considered on the basis of adaptive capacity indicators in the study area. To find solutions to the projected impacts of climate change, the multi-criteria analysis (MCA) was applied following the following three steps.

(a) Defining attributes or parameters for the evaluation of adaptation strategies
Evaluation criteria were used to evaluate the alternative strategies to decide the priority or best adaptation strategy for the study area. The criteria were based on the attributes of efficiency, sustainability and cost. These aspects were considered important as they affect the implementation of adaptation measures; moreover, they are part of the principles of integrated water resources management following the Dublin statement on sustainable management of water resources (Kundzewicz 1997) and the sustainable development goal 13 with targets 13.1, 13.2 and 13.3, which are about adaptation and integration of climate change measures. A typical example is the cost of increasing awareness or changing the cropping system, efficiency in helping to adapt to impacts of climate change (e.g. provision of financial assistance), or the sustainability of crop diversification. The cost, sustainability and efficiency aspects were estimated using the socio-economic status observed in the area and institutional framework in place, pointing to the adaptive capacity indicators for the Kikafu area. Based on their relative importance, these attributes for adaptation were rated in scale (1–3) using field survey data generated from farmers and extension workers as well as expert judgment. The outcome provided a basis for ranking of the attributes in order of importance or relative ability for the implementation of the strategies. The most important is ranked first followed by the next important and so on. These attributes for adaptation were then assigned importance weights to allow for trade-off among different attributes. The weights were obtained using Equation (3) with inputs from the ranking: 
formula
3
where wj is weight of the jth attribute, m is the number of attributes, rj is the rank of the jth attribute and p is the parameter for weights distribution. If p = 0, the weights are equal and as p increases, higher weights are given to highly ranked attributes. Normally, the p parameter is considered to be 1.
(b) Searching for or proposing the adaptation strategies

The proposed strategies are alternatives or alternative strategies, which present a different choice of measures available for use to counter the impacts of climate change on CWRs. These were obtained from literature reviewed and field questionnaire surveys conducted with farmers and agricultural officers in Kikafu River sub-catchment. In all, 20 farmers were interviewed, 10 on the upper side and 10 from the lower side of the Kikafu catchment area. The survey was conducted during the first week of March 2014. The interview focused on the crops grown, cropping pattern, climate pattern, incidences of droughts and crop failures, and adaptation measures taken or those could be taken.

The measures to be put in place should involve both the farmers and the government. Climate change adaptive measures are encouraged at both the local and national scale to ensure food security (McCarthy et al. 2001). The farmers are particularly important because they are characterised by low means of adapting to climate variation impacts (Agrawala et al. 2003). As smallholder farmers operate at the local scale, they will, among others, shift to more drought resistant crops, plant early maturing crops or change their cropping system in response to the projected increase or decrease in temperatures or precipitation. Each of the proposed strategies were also rated using a rating scale (1–3) based on the importance or suitability for the adaptation evaluation criteria (i.e. efficiency, sustainability and cost).

(c) MCA of proposed strategies so that to identify the order of priority and best strategies

A decision matrix was formed based on the results of alternatives and attributes for the adaptation strategies. To compare and rank the adaptation strategies, normalisation was done. The normalised value was obtained from multiplying the weight of the attribute and the rank of the attribute. In this case all the attributes were converted to an equal scale of between 0 and 1. The normalised scores for each strategy considering the three attributes were then summed to get the total score. The total scores were then ranked from the highest to the lowest and this provided the priority of the adaptation strategies.

RESULTS AND DISCUSSION

Calibration and validation of LARS-WG model

The LARS-WG was calibrated and validated using the site analysis and QTest options in the model, respectively. The baseline data observed were used and the results of the statistical analysis inherent in simulating the seasonality of the data observed. The daily rainfall distribution is presented in Tables 4 and 5 for Moshi Airport and Lyamungu station, respectively. The remark column was used to rate the performance of the model with the scales of 1–0.9 for perfect, 0.9–0.7 for good, 0.7–0.5 for satisfaction and below 0.5 for poor. Overall, the K–S test results were perfect since the values (indicating the comparison of probability distributions for each month or season for generated and observed weather) are low and the p-value is more than the significant level of 0.05. These results show that the model can simulate well the wet and dry series at the stations. A wet day is defined as a day with precipitation of >0.0 mm whereas a dry day has 0.0 mm of precipitation. The events of wet day and dry day formed the wet and dry series.

Table 4

K–S test results and p-values for seasonal distribution of wet and dry series

Months Season K–S p-value Remark 
Moshi Airport 
 DJF Wet 12 0.058 Perfect 
 DJF dry 12 0.086 Perfect 
 MAM Wet 12 0.030 Perfect 
 MAM dry 12 0.036 Perfect 
 JJA Wet 12 0.028 Perfect 
 JJA dry 12 0.040 Perfect 
 SON Wet 12 0.030 Perfect 
 SON dry 12 0.071 Perfect 
Lyamungu station 
 DJF Wet 12 0.014 Perfect 
 DJF Dry 12 0.069 Perfect 
 MAM Wet 12 0.030 Perfect 
 MAM Dry 12 0.030 Perfect 
 JJA Wet 12 0.030 Perfect 
 JJA Dry 12 0.044 Perfect 
 SON Wet 12 0.196 0.7203 Good 
 SON dry 12 0.070 Perfect 
Months Season K–S p-value Remark 
Moshi Airport 
 DJF Wet 12 0.058 Perfect 
 DJF dry 12 0.086 Perfect 
 MAM Wet 12 0.030 Perfect 
 MAM dry 12 0.036 Perfect 
 JJA Wet 12 0.028 Perfect 
 JJA dry 12 0.040 Perfect 
 SON Wet 12 0.030 Perfect 
 SON dry 12 0.071 Perfect 
Lyamungu station 
 DJF Wet 12 0.014 Perfect 
 DJF Dry 12 0.069 Perfect 
 MAM Wet 12 0.030 Perfect 
 MAM Dry 12 0.030 Perfect 
 JJA Wet 12 0.030 Perfect 
 JJA Dry 12 0.044 Perfect 
 SON Wet 12 0.196 0.7203 Good 
 SON dry 12 0.070 Perfect 
Table 5

K–S results and p-value for daily rainfall distribution

Month K–S p-value Remark 
Moshi Airport 
 Jan 12 0.1910 0.7494 Good 
 Feb 12 0.0740 1.000 Perfect 
 Mar 12 0.0570 1.000 Perfect 
 Apr 12 0.0670 1.000 Perfect 
 May 12 0.0400 1.000 Perfect 
 Jun 12 0.0970 1.000 Perfect 
 Jul 12 0.0340 1.000 Perfect 
 Aug 12 0.0500 1.000 Perfect 
 Sep 12 0.1700 0.8611 Good 
 Oct 12 0.1150 0.9963 Perfect 
 Nov 12 0.0530 1.000 Perfect 
 Dec 12 0.0430 1.000 Perfect 
Lyamungu station 
 Jan 12 0.165 0.8830 Good 
 Feb 12 0.044 Perfect 
 Mar 12 0.055 Perfect 
 Apr 12 0.070 Perfect 
 May 12 0.056 Perfect 
 Jun 12 0.075 Perfect 
 Jul 12 0.030 Perfect 
 Aug 12 0.089 Perfect 
 Sep 12 0.870 Perfect 
 Oct 12 0.059 Perfect 
 Nov 12 0.091 Perfect 
 Dec 12 0.053 Perfect 
Month K–S p-value Remark 
Moshi Airport 
 Jan 12 0.1910 0.7494 Good 
 Feb 12 0.0740 1.000 Perfect 
 Mar 12 0.0570 1.000 Perfect 
 Apr 12 0.0670 1.000 Perfect 
 May 12 0.0400 1.000 Perfect 
 Jun 12 0.0970 1.000 Perfect 
 Jul 12 0.0340 1.000 Perfect 
 Aug 12 0.0500 1.000 Perfect 
 Sep 12 0.1700 0.8611 Good 
 Oct 12 0.1150 0.9963 Perfect 
 Nov 12 0.0530 1.000 Perfect 
 Dec 12 0.0430 1.000 Perfect 
Lyamungu station 
 Jan 12 0.165 0.8830 Good 
 Feb 12 0.044 Perfect 
 Mar 12 0.055 Perfect 
 Apr 12 0.070 Perfect 
 May 12 0.056 Perfect 
 Jun 12 0.075 Perfect 
 Jul 12 0.030 Perfect 
 Aug 12 0.089 Perfect 
 Sep 12 0.870 Perfect 
 Oct 12 0.059 Perfect 
 Nov 12 0.091 Perfect 
 Dec 12 0.053 Perfect 

A comparison of the statistical parameters of the simulated data with those calculated from the data observed (Figure 3(a) and 3(b)) indicates that the model has perfect performance in modelling the mean monthly totals. It models well the mean monthly totals, with the standard deviation also modelling well except for the months of April and November for Moshi Airport for which the model seems to underestimate the standard deviation. For Lyamungu station, there are large differences in standard deviation for March, April and May. Such differences could mainly be attributed to the model smoothing the data observed (Semenov & Barrow 2002). Smoothing eliminates as much as possible the random noise in the data observed to get closer to the actual climate for the site (Semenov & Barrow 2002). Such differences are likely to be caused by departures of the values observed from the smooth pattern for the data and can be occasioned by errors in the data observed, random variations in data observed and climate anomalies (Semenov & Barrow 2002).
Figure 3

Observed (obs) and simulated (gen) rainfall mean monthly totals and standard deviation (sd) during 1971–2000 for (a) Moshi Airport and (b) Lyamungu station.

Figure 3

Observed (obs) and simulated (gen) rainfall mean monthly totals and standard deviation (sd) during 1971–2000 for (a) Moshi Airport and (b) Lyamungu station.

Downscaled future climate data

Temperature projection

The temperature changes from the three GCMs show an increase for all the months (Figure 4). The models indicate temperature increases in the period of 2011–2030 ranging from 0.61 to 1.18°C with the GFCM21 model recording the highest increase of 1.18°C in the month of May while HADCM3 shows the lowest of 0.6°C in March. In the period of 2046–2065, temperature change ranges from 1.41 to 2.7°C. Higher temperature increases are expected in the months of May, June and July for HADCM3 and GFCM21 models. The CNCM3 model indicates the highest temperature increases in the months of July, August and September and the lowest for April and May. The United Nations Development Programme (UNDP) climate change country projections for temperature show that the mean annual temperature is projected to increase by 1–2.7°C by 2060 and 1.5–4.5°C in 2090 (McSweeney et al. 2010). The results above show that the temperature increase projected by the GCMs is well within the range projected by UNDP.
Figure 4

Projected monthly changes in temperature for three GCMs for A2 climate scenario at Moshi Airport.

Figure 4

Projected monthly changes in temperature for three GCMs for A2 climate scenario at Moshi Airport.

Precipitation projection

In the precipitation projections, unlike for temperature, the three models show inconsistencies in the precipitation changes (Figures 5 and 6). This can be partly due to the models’ inability to capture orographic influences (Hudson & Jones 2002) and the difference in climate sensitivity due to parameterisation and simplification of processes (Barrow et al. 2004).
Figure 5

Precipitation change factors for (a) 2020s and (b) 2050s for the three GCMs at Moshi Airport.

Figure 5

Precipitation change factors for (a) 2020s and (b) 2050s for the three GCMs at Moshi Airport.

Figure 6

Precipitation change factors for (a) 2020s and (b) 2050s for the three GCMs at Lyamungu station.

Figure 6

Precipitation change factors for (a) 2020s and (b) 2050s for the three GCMs at Lyamungu station.

Seasonal variability of precipitation changes

From the estimated precipitation scenarios at Moshi Airport, Figure 7(a) and 7(b) shows a projected increase of up to 35% or a decrease of 19% in the 2020s. In the 2050s, a 56% precipitation increase is projected whereas the decrease will be at a minimum of 25% for the CNCM3 and HADCM3 models. However, GFCM21 model shows a very high precipitation increase of 205% and a minimum decrease of 10% for the 2050s.
Figure 7

Percent changes in monthly precipitation for (a) 2020s and (b) 2050s at Moshi Airport.

Figure 7

Percent changes in monthly precipitation for (a) 2020s and (b) 2050s at Moshi Airport.

Climate change impacts on CWRs

Trend analysis of rainfall and temperature

The performance of the linear trend analysis for the annual mean minimum and mean maximum temperature shows evidence of an increasing trend that hints at changing climate in the sub-catchment area. The rates of increase in both maximum (Tmax) and minimum (Tmin) temperature for the Moshi Airport are 0.01 and 0.02, respectively, with 1983 and 1987 having the highest maximum temperatures of 30.1 and 30.3°C (Figure 8(a) and 8(b)), respectively, whereas the lowest minimum temperature for the stations is 12.4°C in 1971 (Figure 9(a) and 9(b)). The trend significance from the analysis using Mann–Kendall shows that Tmax and Tmin had a statistically significant upswing trend for both Moshi Airport and Lyamungu station. The rates of increase in both Tmax and Tmin temperature for the Moshi Airport are 0.018°C/year and 0.024°C/year at α = 0.1 and α = 0.01, respectively. The rates of increase in both Tmax and Tmin temperature for Lyamungu are 0.017°C/year and 0.067°C/year at α = 0.05 and α = 0.001, respectively.
Figure 8

Linear trend for maximum temperature at (a) Moshi Airport and (b) Lyamungu station.

Figure 8

Linear trend for maximum temperature at (a) Moshi Airport and (b) Lyamungu station.

Figure 9

Linear trend for minimum temperature at (a) Moshi Airport and (b) Lyamungu station.

Figure 9

Linear trend for minimum temperature at (a) Moshi Airport and (b) Lyamungu station.

The annual rainfall in the catchment area shows a decreasing trend for Moshi Airport and Lyamungu, respectively (Figure 10). However, the trend was not statistically significant at α = 0.1, 0.05, 0.01 and 0.001. The highest annual rainfall of 1356.5 mm at Moshi Airport was received in 1979 and the lowest of 511.9 mm was in 1973. Lyamungu station received a high of 2124.7 mm in 1990 and a low of 924.2 mm in 1987. This difference in occurrences of extreme rainfall indicates spatial variation of rainfall in and around the sub-catchment. The trend of more or less the same rainfall and increasing temperature is expected to impact on the CWRs as it affects the water availability to the crop. The increase in temperature increases the ETo and consequently CWR.
Figure 10

Linear trend of rainfall for Moshi Airport and Lyamungu station.

Figure 10

Linear trend of rainfall for Moshi Airport and Lyamungu station.

The trend analysis using Mann–Kendall reveals that the increase in maximum and minimum temperature was significant whereas for precipitation, the decrease was insignificant. With the significant increase in temperature, the reference evapotranspiration increases will lead to increases in the CWRs.

Crop water requirements

The maize CWR for the baseline and future scenarios were calculated for the entire growing season (see Tables 6 and 7). The average CWR in the baseline period was 615.9 mm per year for Moshi Airport and 472 mm per year for Lyamungu station. The months of April and May have the highest CWRs and August has the lowest at 20.7 mm. For future periods, the results showed an increase in CWRs from 625.2–639 mm per year in the 2020s to 646–659.9 mm per year in the 2050s. For the 2020s, the CWRs increased by 1.5–3.8% whereas in the 2050s the increase was between 4.9 and 7.1%. Lyamungu station has a higher increase in CWR with the 2020s having an increase of 19.9% and the 2050s having 22.4%.

Table 6

Projected CWRs for three GCMs at Moshi Airport

Time period Baseline (mm) CNCM3 (mm) GFCM21 (mm) HADCM3 (mm) 
1971–2000 615.9    
ETc-2020s  625.2 639 625.6 
ETc-2050s  646 648.4 659.9 
% increase 2020s  1.5 3.8 1.57 
% increase 2050s  4.9 5.3 7.1 
Time period Baseline (mm) CNCM3 (mm) GFCM21 (mm) HADCM3 (mm) 
1971–2000 615.9    
ETc-2020s  625.2 639 625.6 
ETc-2050s  646 648.4 659.9 
% increase 2020s  1.5 3.8 1.57 
% increase 2050s  4.9 5.3 7.1 
Table 7

Projected CWRs for three GCMs at Lyamungu station

Time period Baseline (mm) CNCM3 (mm) GFCM21 (mm) HADCM3 (mm) 
1971–2000 472    
ETc-2020s  557.8 566 558.5 
ETc-2050s  575.7 577.5 577.9 
% increase 2020s  18 19.9 18.3 
% increase 2050s  21.9 22.4 22.4 
Time period Baseline (mm) CNCM3 (mm) GFCM21 (mm) HADCM3 (mm) 
1971–2000 472    
ETc-2020s  557.8 566 558.5 
ETc-2050s  575.7 577.5 577.9 
% increase 2020s  18 19.9 18.3 
% increase 2050s  21.9 22.4 22.4 

The CWRs for the entire growing season starting from 3rd March to 29th August showed increasing decadal (dec) CWRs, especially for the development and mid-season stages (see Figures 11 and 12).
Figure 11

CWR for 2020s relative to the baseline.

Figure 11

CWR for 2020s relative to the baseline.

Figure 12

CWR for 2050s relative to the baseline.

Figure 12

CWR for 2050s relative to the baseline.

Maize appears to be relatively tolerant to water deficits during the initial and late season periods. The greatest decrease in grain yields is caused by water deficits during the development period including tasselling and silking, and pollination due mainly to a reduction in grain number per cob (FAO 2013). This effect is less pronounced when in the preceding initial period the plant has suffered water deficits. Severe water deficits during the development period, particularly at the time of silking and pollination, may result into little or no grain yield due to silk drying. Water deficits during the mid-season period may lead to reduced yield due to a reduction in grain size. On the other hand, water deficit during the late season period has little effect on grain yield (FAO 2013).

From the future projection of CWRs for Moshi Airport (Figures 13 and 14), the last part of the development stage (∼70–80 days) and the entire mid-season stage will experience an increase in CWR that may lead to reduced yields due to the reduction in grain sizes if these requirements are not satisfied by rainfall. From the calculated CWRs and the effective rainfall for the growing stages, the water available to the maize during the last part of the development stage and the mid-season stage will be less than the CWR (Figures 1315). This will have an effect on the grain sizes and will lead to reduced yields. The projection of precipitation shows that the precipitation will decrease in the months of May and June, which are critical months for maize growth. This has the effect of increasing the deficiency of meeting the CWRs in those critical growth stages.
Figure 13

CWR calculated from downscaled CNCM3 for different crop stages for the 2020s and the 2050s.

Figure 13

CWR calculated from downscaled CNCM3 for different crop stages for the 2020s and the 2050s.

Figure 14

CWR calculated from downscaled GFCM21 for the different crop stages for the 2020s and the 2050s.

Figure 14

CWR calculated from downscaled GFCM21 for the different crop stages for the 2020s and the 2050s.

Figure 15

CWR calculated from downscaled HADCM3 for the different crop stages for the 2020s and the 2050s.

Figure 15

CWR calculated from downscaled HADCM3 for the different crop stages for the 2020s and the 2050s.

Assessment of adaptation strategies

From the climate change results it was evident that the temperature in Kikafu is rising and is increasing the CWRs. However, the precipitation changes were not statistically revealing any trend. This points to an increase in the crop water demand at a time when there is a decrease in crop water supply. It is, therefore, important to put in place some adaptation measures so as to help farmers improve food production in line with future scenarios.

The field survey produced the following results: In relation to climate change, the farmers were experiencing variability in the amounts of rainfall they received in the area. Over the years, they sometimes experienced a decline in the amount of rainfall they received and even experienced more rainfall over the past five years. The farmers in the lower areas have experienced drought conditions, especially in the past five years. Some farmers had even considered not planting maize due to rainfall failure and many crop failures. The farmers also faced problems during the onset of the rainy season, either coming early or slightly late in some years which affected the crop growth. The farmers also indicated the rise in temperature levels. In relation to adaptation capacity and strategies, most of the farmers in the study area are poor and therefore their adaptive capacity is very low. The farmers also have very little information on how to adapt adequately to the impacts of the changing climate in the area in addition to lacking requisite skills.

As such, adaptation to impacts of climate change operates at two levels: farm and government levels. Using the results from future climate scenarios and their impacts on CWRs, field survey and literature, the following alternative strategies were obtained: increasing awareness and knowledge, strengthening institutions, provision of financial assistance, crop diversification, changes in cropping system and restoration of organic soils. These are further discussed below.

Adaptation at farm level

Crop diversification

Farmers are encouraged to increase the variety of crops they plant so as to move towards the more drought-resistant ones. They should also plant fast-maturing crops. Drought-resistant crops can withstand the increasing CWRs and, therefore, maximise crop production. The farmers can also plant maize that has shorter growth stages. This can ensure that the two critical stages of development and mid-season are well into the period where effective rainfall can meet the CWRs.

Change in cropping system

The farmers can also change their cropping system to help improve efficiency in crop production. The increasing temperatures increase the CWRs. Therefore, the use of conservation tillage and rotation can help improve soil properties, increase soil water retention capacity and boost the quality of produces.

Restoration of soil organic matter

Farmers are encouraged to use organic manure from livestock and practice mulching. Doing so would help improve the soil texture, increase the soil water retention capacity and hence lead to greater soil water conservation, resilience and increased yields.

Adaptation at government level

The government plays a major role in mainstreaming the adaptations to the changing climate. It is mainly involved in policy-making and implementations on a national or regional scale. Their input also plays a major role in aiding implementation at the farm levels. This can be done by the following.

Increasing awareness and knowledge

For effective adaptation, the government should carry out campaigns and public forum initiatives for sensitising farmers on the impacts of climate change and CWRs and how they can adapt to the changes in their respective areas. Farmers can be taught the best conservation practices and alternative crops that can flourish under increasing crop water demands.

Strengthening institutions

The government should strengthen institutions involved and affected by climate change. In this regard, zonal and district agricultural offices should be capacitated and equipped to handle the increasing CWRs in the sub-catchment area.

Providing financial assistance and incentives

The implementation of adaptation measures requires a great deal of financial resources. The government in this regard needs to provide financial assistance to farmers seeking to adopt irrigated agricultural practices during the crop growth stages when the CWRs are not being met. The government should also initiate incentives that encourage farmers to change their agricultural activities, such as supplying farmers with drought-resistant seeds and fast-maturing crop varieties.

Analysis and identification of the best adaptation strategy

To get a prioritisation and the best suitable strategy for the adaptation of the impacts of climate change on maize CWRs, MCA was used. For the MCA, the final scores were obtained by ranking, weighting using Equation (3) and finally normalisation. The strategies were ranked using the attributes of cost, efficiency and sustainability (Table 8). Normalisation was achieved by multiplying the corresponding ranks and weights (Table 9).

Table 8

Ranking of proposed adaptation strategies

Evaluation attributes/strategies Efficiency Cost Sustainability 
Increasing awareness and knowledge 
Strengthening institutions 
Providing financial assistance 
Crop diversification 
Changes in cropping system 
Restoration of organic soils 
Evaluation attributes/strategies Efficiency Cost Sustainability 
Increasing awareness and knowledge 
Strengthening institutions 
Providing financial assistance 
Crop diversification 
Changes in cropping system 
Restoration of organic soils 
Table 9

MCA of adaptation strategies

Evaluation parameters/strategies Efficiency Cost Sustainability Outcome Priority 
Increasing awareness and knowledge 0.17 0.99 2.16 
Strengthening institutions 1.5 0.17 0.66 2.33 
Providing financial assistance 0.5 0.51 0.66 1.67 
Crop diversification 1.5 0.17 0.66 2.33 
Changes in cropping system 0.5 0.34 0.99 1.83 
Restoration of organic soils 0.17 0.99 2.16 
Evaluation parameters/strategies Efficiency Cost Sustainability Outcome Priority 
Increasing awareness and knowledge 0.17 0.99 2.16 
Strengthening institutions 1.5 0.17 0.66 2.33 
Providing financial assistance 0.5 0.51 0.66 1.67 
Crop diversification 1.5 0.17 0.66 2.33 
Changes in cropping system 0.5 0.34 0.99 1.83 
Restoration of organic soils 0.17 0.99 2.16 

The analysis shows that the highest priority was given to crop diversification and strengthening of institutions followed by restoration of soil organic matter and increasing awareness and knowledge. This was followed by changing of cropping systems and subsequently providing financial assistance.

CONCLUSIONS AND RECOMMENDATIONS

This research has established future climate change scenarios and their impacts on maize CWRs, and prioritised adaptation strategies for Kikafu sub-catchment. The temperature was projected to increase, with the 2020s period having an increase of 1.18°C and 2050s increase of 2.7°C. From HADCM3 and GFCM21 models, high temperature increases are expected in the months of May, June and July whereas temperature decreases are projected to be experienced during the months of March and April. Precipitation projections, however, show variability with the increase expected in some months with others experiencing a decrease. The three GCMs data used in the study also reveal variability in their projections. For example, precipitation increase is expected especially in the months of February, March and April and decreases in the months of June, July and August. From the CNCM3 and HADCM3 models, in the 2020s and 2050s, an increase is projected to stand at 35 and 56% and the decreases at 19 and 25%, respectively. The GFCM21 model, however, shows a higher increase of 205% in the 2050s.

The analysis of impacts of climate change on CWR shows an increase in both the 2020s and the 2050s relative to the baseline period of 1971–2000. There is a higher projected increase at Lyamungu station than at Moshi Airport. In the 2020s, Moshi Airport is expected to have a maximum increase of 3.8% whereas in the 2050s it will be 7.1%. Lyamungu station is expected to have a maximum increase of 19.9% in the 2020s and 22.4% in the 2050s. Analysis of CWRs during the crop growth stages shows that there will be high unmet CWRs that might affect the yields, especially in the development and mid-season stages. In consequence, more broad adaptation strategies to these impacts are important and necessary. Based on the MCA, the study shows that strengthening of institutions and crop diversification, increasing awareness and knowledge, and restoration of soil organic matter were among priority adaptation strategies in response to climate change impacts on maize crop. Also, future research may be required on crops and varieties which are water-efficient and fast-maturing so as to avoid irrigation water requirements.

The findings of this study can be used by district agricultural offices and by the relevant ministries at the policy level for agricultural planning in the face of climate change. However, for a complete analysis of the agroclimatic characteristics or rain-fed agricultural practices in the sub-catchment, it is necessary to establish variation or change of the onset and cessation periods for the rainy seasons and the duration of the rainy season, which can be linked with the variation or change of the CWRs. Also, the study used three GCMs and A2 emission scenario. Since climate change is dynamic and cannot be understood fully or accurately predicted, the study can be extended further by using more GCMs and emission scenarios to get a wider range of uncertainty bound.

ACKNOWLEDGEMENTS

The Master's Degree programme scholarship to the first author from the Southern Africa Development Community (SADC)’s WaterNet is hereby highly appreciated. We are also grateful to the SSI-2 project for provision of daily processed climatic data for this study.

REFERENCES

REFERENCES
Agrawala
S.
Moehner
A.
Hemp
A.
Aalst
M.
Hitz
S.
Smith
J.
Meena
H.
Mwakifwamba
S.
Hyera
T.
Mwaipopo
O.
2003
Development and climate change in Tanzania: Focus on Mount Kilimanjaro
.
Int. J. Mod. Soc. Sci.
2
(
3
),
169
194
.
Allen
R. G.
Pereira
L. S.
Raes
D.
Smith
M.
1998
Crop Evapotranspiration. Guidelines for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper No. 56
.
Food and Agriculture Organization
,
Rome
,
Italy
.
AUC-AMCOW
2016
African Water Resources Management Priority Action Programme 2016–2025 (WRM-PAP)
.
African Union Commission (AUC) and African Minister's Council on Water (AMCOW)
.
Barrow
E. M.
Maxwell
B.
Gachon
P.
2004
Climate Variability and Change in Canada: Past, Present and Future, Climate Change Impacts Scenarios Project. National Report
,
Environment Canada, Meteorological Service of Canada, Adaptation Impacts Research Group, Atmospheric and Climate Sciences Directorate Publication
,
Canada
, pp.
114
.
Eden
J. M.
Widmann
M.
Grawe
D.
Rast
S.
2012
Skill, correction, and downscaling of GCM-simulated precipitation
.
J. Clim.
25
,
3970
3984
.
FAO
1998
Crop Evapotranspiration. Guidelines for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper No. 56
.
Food and Agriculture Organization
,
Rome
,
Italy
.
FAO
2013
The State of Food Insecurity in the World
.
Food and Agriculture Organization
,
Rome
,
Italy
.
FAO
2015
Crop water information: Maize. FAO WATER. Land and Water Division. Available from: www.fao.org/nr/water/cropinfo_maize.html (accessed 28 January 2016)
.
FAO
2016
Revised FAO Methodology for Crop Water Requirements. Available from: www.fao.org/nr/water/docs/revised-fao-methodology-cropwaterrequirements.pdf (accessed 28 June 2016)
.
Gulacha
M. M.
Mulungu
D. M. M.
2016
(In press)
Generation of climate change scenarios for precipitation and temperature at local scales using SDSM in Wami-Ruvu River basin Tanzania
.
J. Phys. Chem. Earth (JPCE)
.
DOI:10.1016/j.pce.2016.10.003.
Hirsch
R. M.
Slack
J. R.
Smith
R. A.
1982
Techniques of trend analysis for monthly water quality data
.
Water Resour. Res.
18
,
107
121
.
Holmén
H.
2003
Reflections on Natural Reconditions for a Green Revolution in Africa
.
Department of Geography, Linkoping University, The Tema Institute
,
SE-581 83
,
Linkoping
.
Hudson
D. A.
Jones
R.
2002
Simulations of Present-day and Future Climates of Southern Africa Using HadAM3H
.
Hadley Centre for Climate Prediction and Research, Meteorological Office
,
Bracknell
,
UK
.
Hulme
M.
(ed.)
1996
Climate Change and Southern Africa: An Exploration of Some Potential Impacts and Implications in the SADC Region
.
Climatic Research Unit, University of East Anglia
,
Norwich
,
UK
and WWF International, Gland, Switzerland
, p.
104
.
Iizumi
T.
Semenov
M. A.
Nishimori
M.
Ishigooka
Y.
Kuwagata
T.
2012
ELPIS-JP, a dataset of local-scale daily climate change scenarios for Japan
.
Philos. Trans. R Soc. A
370
,
1121
1139
.
IPCC
2007
Mitigation of Climate Change, Technical Summary Report, AR4
.
Intergovernmental Panel on Climate Change (IPCC)
,
Geneva
.
IPCC-TGICA
2007
General Guidelines on the Use of Scenario Data for Climate Impact and Adaptation Assessment. Version 2. Prepared by T. R. Carter on behalf of the Intergovernmental Panel on Climate Change, Task Group on Data and Scenario Support for Impact and Climate Assessment
, p.
66
.
IUCN
2010
Climate change adaptation in Pangani River Basin brief
.
J. Mod. Social Sci.
2
(
3
),
169
194
.
Kundzewicz
Z. W.
1997
Water resources for sustainable development
.
Hydrol. Sci. J.
42
(
4
),
467
480
.
Kundzewicz
Z. W.
Robson
A.
(eds)
2000
Detecting Trend and Other Changes in Hydrological Data
.
World Climate Program – Water, WMO/UNESCO, WCDMP45, WMO/TD 1013
,
Geneva
, p.
157
.
Liwenga
E. T.
Kangalawe
R. Y. M.
Lyimo
J. G.
Majule
A. E.
Ngana
J. O.
2007
Research Protocols for Assessing the Impact of CC & V in Rural Tanzania: Water, Food Systems, Vulnerability and Adaptation
.
START/PACOM, African Global Change Research
,
Washington, DC
,
USA
.
McCarthy
J. J.
Canziani
O. F.
Leary
N. A.
Dokken
D. J.
White
K. S.
(eds)
2001
Climate Change 2001: Impacts, Adaptation, and Vulnerability
.
Cambridge University Press
,
Cambridge
.
Morton
J.
2007
The impact of climate change on smallholder and subsistence agriculture
.
Proc. Natl. Acad. Sci. USA
104
,
50
.
Mulungu
D. M. M.
2014
Water Processes at Different Spatial Scales: The Case of Upper Pangani River Basin (Kikuletwa and Rau River Sub-catchments). A draft research report for Up-scaling Small-Scale Land and Water Systems Innovations in Dryland Agro-ecosystems for Sustainability and Livelihood Improvements (SSI-2) Project. Department of Water Resources Engineering, University of Dar es Salaam
.
NAPA
2006
National Adaptation Programme of Action 2006. Government Programme Document
,
United Republic of Tanzania
.
Noufea
D.
Mahea
G.
Kamagatea
E.
Servat
A.
Goula
B. T.
Savanae
I.
2015
Climate change impact on agricultural production: the case of Comoe River basin in Ivory Coast
.
Hydrol. Sci. J.
60
(
11
),
1972
1983
.
Ravindranath
N. H.
Manuvie
R.
Lakshmi
C. S.
2010
Biofuels and climate change
. In:
Food Versus Fuel
(
Rosillo-Calle
F.
Johnson
F. X.
, eds).
Zed Books Ltd
,
London
and New York
, pp.
217
.
Rehana
S.
Mujumdar
P. P.
2013
Regional impacts of climate change on irrigation water demands
.
Hydrol. Process.
27
(
20
),
2918
2933
.
Rohr
P. C.
2003
A hydrological study concerning the southern slopes of Mt. Kilimanjaro, Tanzania
.
Doctoral dissertation
,
Department of Hydraulic and Environmental Engineering, Faculty of Engineering Science and Technology, Norwegian University of Science and Technology
,
Trondheim
,
Norway
.
Rosenzweig
C.
Tubiello
F. N.
Goldberg
R.
Mills
E.
Bloomfield
J.
2002
Increased crop damage in the U.S. from excess precipitation under climate change
.
Glob. Environ. Change
12
(
3
),
197
202
.
Savva
A. P.
Frenken
K.
2002
Crop Water Requirements and Irrigation Scheduling. Irrigation Manual Module 4
.
FAO Sub-Regional Office for East and Southern Africa
,
Harare
,
Zimbabwe
.
Semenov
M. A.
Barrow
E. M.
2002
LARS-WG: A Stochastic Weather Generator for Use in Climate Impact Studies, Version 3.0, User's Manual
.
Rothamsted Research
,
Harpenden, Hertfordshire
,
UK
.
Semenov
M. A.
Brooks
R. J.
Barrow
E. M.
Richardson
C. W.
1998
Comparison of the WGEN and LARS-WG stochastic weather generators in diverse climates
.
Clim. Res.
10
,
95
107
.
Smith
M.
1991
CROPWAT Manual and Guidelines
.
FAO of UN
,
Rome
,
Italy
.
Smith
J. B.
Hulme
M.
1998
Climate change scenarios
. In:
UNEP Handbook on Methods for Climate Change Impact Assessment and Adaptation Studies
(
Feenstra
J. F.
Burton
I.
Smith
J. B.
Tol
R. S. J.
, eds).
United Nations Environmental Programme, Nairobi, Kenya and Institute for Environmental Studies
,
Amsterdam
, pp.
3-1
3-40
.
Smith
M.
Kivumbi
D.
2006
Calculation Procedure Use of the FAO CROPWAT Model in Deficit Irrigation Studies
.
FAO (Food and Agriculture Organization)
,
Rome
,
Italy
;
Heng, L.K., Joint FAO/IAEA Division, International Atomic Energy Agency, Vienna, Austria
.
Solomon
S.
Qin
D.
Manning
M.
Chen
Z.
Marquis
M.
Averyt
K. B.
Tignor
M.
Miller
H. L.
2007
Climate Change, The Physical Science Basis Contributions of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge
.
Tadross
M.
Wolski
P.
2010
Climate Change Modeling for the Pangani Basin to Support Integrated Water Management Planning Process (Final Report)
.
Climate Systems Analysis Group
,
SA
.
Tan
M. L.
Ibrahim
A. L.
Yusop
Z.
Duan
Z.
Ling
L.
2015
Impacts of land-use and climate variability on hydrological components in the Johor River basin, Malaysia
.
Hydrol. Sci. J.
60
(
5
),
873
889
.
Tubiello
F. N.
Donatelli
M.
Rosenzweig
C.
Stockle
C. O.
2000
Effects of climate change and elevated CO2 on cropping systems: Model predictions at two Italian locations
.
Eur. J. Agron.
12
,
179
189
.
URT
2006
Agricultural Sector Development Programme (ASDP)
.
Government Programme Document
,
United Republic of Tanzania (URT)
.
Wilby
R. L.
Dawson
C. W.
Barrow
E. M.
2002
SDSM – a decision support tool for the assessment of regional climate change impacts
.
Environ. Model. Softw.
17
,
147
159
.
World Meteorological Organization (WMO)
1988
Analyzing Long Time Series of Hydrological Data with Respect to Climate Variability. WCAP-3, WMO/TD No. 224
.
World Meteorological Organization
,
Geneva
,
Switzerland
.